A meningococcal vaccine antigen engineered toincrease thermal stability and stabilizeprotective epitopesMonica Konar, Rolando Pajon1, and Peter T. Beernink2

Center for Immunobiology and Vaccine Development, Children’s Hospital Oakland Research Institute, University of California San Francisco BenioffChildren’s Hospital Oakland, Oakland, CA 94609

Edited by Arie van der Ende, Academic Medical Center, Center for Infection and Immunity Amsterdam (CINIMA), Amsterdam, The Netherlands, and acceptedby the Editorial Board October 15, 2015 (received for review April 21, 2015)

Factor H binding protein (FHbp) is part of two vaccines recentlylicensed for prevention of sepsis and meningitis caused byserogroup B meningococci. FHbp is classified in three phylogenicvariant groups that have limited antigenic cross-reactivity, andFHbp variants in one of the groups have low thermal stability. Inthe present study, we replaced two amino acid residues, R130 andD133, in a stable FHbp variant with their counterparts (L and G)from a less stable variant. The single and double mutants decreasedthermal stability of the amino- (N-) terminal domain compared withthe wild-type protein as measured by scanning calorimetry. We in-troduced the converse substitutions, L130R and G133D, in a less stablewild-type FHbp variant, which increased the transition midpoint (Tm)for the N-terminal domain by 8 and 12 °C; together the substitutionsincreased the Tm by 21 °C. We determined the crystal structure of thedouble mutant FHbp to 1.6 Å resolution, which showed that R130 andD133 mediated multiple electrostatic interactions. Monoclonal anti-bodies specific for FHbp epitopes in the N-terminal domain had higherbinding affinity for the recombinant double mutant by surfaceplasmon resonance and to the mutant expressed on meningococciby flow cytometry. The double mutant also had decreased bindingof human complement Factor H, which in previous studies increasedthe protective antibody responses. The stabilized mutant FHbp thushas the potential to stabilize protective epitopes and increase theprotective antibody responses to recombinant FHbp vaccines or na-tive outer membrane vesicle vaccines with overexpressed FHbp.

Neisseria meningitidis | complement Factor H | site-specific mutagenesis |calorimetry | crystal structure

Factor H binding protein (FHbp) is a key antigen in two multi-component vaccines that have been licensed in the United

States and/or the European Union for prevention of bacterialmeningitis and sepsis caused by Neisseria meningitidis. FHbp is animportant virulence factor, as it binds human Factor H (FH) (1)to the bacterial surface, which negatively regulates complementactivation. FHbp is under immune selection (2) to preserve thisvirulence mechanism and thus exhibits a high degree of antigenicvariation. More than 800 amino acid sequence variants have beenidentified to date (available in the public database at pubmlst.org/neisseria/fHbp). The individual sequence variants can be classifiedin two subfamilies (3) or three variant groups (4) based on aminoacid sequence identity. Antibodies to FHbp from variant group 1 donot elicit complement-mediated bactericidal activity against strainswith FHbp from variant groups 2 or 3 and vice versa; therefore, ithas been proposed that two or more divergent FHbp antigens areneeded to achieve broad protective immunity (3, 4).One of the two licensed vaccines contains an FHbp sequence

from variant groups 1 and 3 and was shown to elicit broad pro-tective immunity against genetically diverse meningococcal strains(3, 5, 6). The second licensed vaccine contains FHbp in variantgroup 1 and three other antigens capable of eliciting protectiveantibodies (7, 8). Other strategies to increase the breadth of

protection were to engineer FHbp antigens containing epitopesfrom different sequence variants (9–11).FHbp sequence variants in variant group 2 are prevalent

among invasive strains in Europe (12) and North America (ID19) (13) and Africa (ID 22) (14). These two sequence variantswere reported to have low thermal stability for the amino- (N-)terminal domain (15, 16), whereas FHbp from variant groups 1and 3 had high thermal stability (15, 17, 18). The relatively lowthermal stability for FHbp variant group 2 proteins was sug-gested to limit the vaccine potential of these variants (15). In thepresent study, we used sequence alignments and the known 3Dstructure of FHbp (19) to identify two amino acid residues inFHbp variant group 1 that were important for thermal stability.Using this information, we engineered a mutant FHbp in variantgroup 2 that had increased thermal stability and determined itsstructure at high resolution. The mutant antigen had increasedaffinity for protective monoclonal antibodies (MAbs) and has thepotential to improve next-generation FHbp vaccines.

ResultsThermal Stability of Natural FHbp Sequence Variants.A stable FHbpvariant (ID 1 in variant group 1) unfolded with two thermal tran-sitions (Fig. 1A), which corresponded to the N- and carboxyl- (C-)terminal domains (17). The transitions had midpoint (Tm) values of

Significance

Factor H binding protein (FHbp) is a component of two vaccinesrecently licensed for prevention of sepsis and meningitis causedby meningococci. FHbp is antigenically variable, and certain se-quence variants have low thermal stability. Two amino acid sub-stitutions stabilized a less stable FHbp variant by 21 °C, and thehigh-resolution crystal structure of the stabilized FHbp antigenshowed that the two new residues mediated multiple electro-static interactions. The two substitutions increased the affinity formonoclonal antibodies specific for different regions of FHbp andincreased binding to the mutant expressed on the surface ofmeningococci. The stabilized FHbp antigen has the potential toimprove the immunogenicity of meningococcal vaccines contain-ing recombinant FHbp or native outer membrane vesicles.

Author contributions: M.K., R.P., and P.T.B. designed research; M.K., R.P., and P.T.B. per-formed research; M.K. and P.T.B. analyzed data; and P.T.B. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission. A.v.d.E. is a guest editor invited by the EditorialBoard.

Data deposition: The crystallography, atomic coordinates, and structure factors have beendeposited in the Protein Data Bank, www.pdb.org (PDB ID code 4Z3T).1Present address: Vaccine and Novel Immunotherapy Solutions, Covance Inc., Indianapo-lis, IN 46214.

2To whom correspondence should be addressed. Email: pbeernink@chori.org.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1507829112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1507829112 PNAS | December 1, 2015 | vol. 112 | no. 48 | 14823–14828

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Mar

ch 7

, 202

0

69.1 and 84.7 °C (Table S1). A less stable FHbp variant (ID 22 invariant group 2) also unfolded with two transitions (Fig. 1D), butwith Tm values of 38.5 and 82.3 °C (Table S1). The enthalpy change(ΔH) for the N-terminal domain of the variant group 2 protein alsowas lower than the variant group 1 protein (12 and 96 kcal/mol).Whereas the Tm and ΔH values for the C-terminal domain weresimilar for the two proteins, the stability of the N-terminal domainin the variant group 2 protein was lower, both in terms of transitiontemperature (ΔTm = 30.6 °C) and enthalpy (ΔΔH = 83 kcal/mol).To assess whether the thermal stability of the N-terminal do-

main was low for other variant group 2 proteins, we examinedtwo additional sequence variants, ID 19 and 77 (Fig. S1). FHbpID 19 had a Tm value of 40.6 °C and a ΔH of 16 kcal/mol forunfolding of the N-terminal domain. FHbp ID 77 had a Tm valueof 57.8 °C and a ΔH of 34 kcal/mol. Although the stability of ID77 was higher than ID 22, the stability of ID 77 was low com-pared with ID 1 in variant group 1 (Table S1). Using a multiplesequence alignment of three FHbp amino acid sequences fromeach of variant groups 1 and 2, we identified 25 differences withinthe N-terminal domain (Fig. S2). Of these, seven were non-conservative differences. Based on the locations and interactionsof these seven residues in the crystal structure of FHbp (19), weidentified two positions, 130 and 133, that we hypothesized toconfer stability in the variant group 1 proteins.

Reciprocal Substitutions in FHbp Variant Groups 1 and 2. To testwhether these two amino acid residues contributed to the stability ofFHbp, we replaced each of the two residues in a stable FHbpvariant (ID 1 in variant group 1) with the corresponding residuefrom less stable variants in variant group 2. The R130L sub-stitution in FHbp ID 1 decreased the Tm value of the N-terminaldomain by 4.0 °C, whereas there was little effect on the C-ter-minal domain (Fig. 1A and Table S1). The D133G substitutiondecreased the Tm of the N-terminal domain by 10.4 °C (Fig. 1B).The R130L/D133G double mutant showed an effect that wasslightly more than additive, with a Tm that was 16.3 °C lower thanthe wild-type (WT) protein (Fig. 1C).We further reasoned that the converse amino acid residue

changes might stabilize FHbp in variant group 2. We chose FHbpID 22, as this sequence variant was prevalent in invasive isolatesfrom Africa (14, 20) and the immunogenicity of vaccines con-taining FHbp ID 22 previously had been tested in mice (6, 14, 16,21). The L130R substitution increased the Tm of the N-terminaldomain by 7.7 °C (Fig. 1D). The G133D substitution increasedthe Tm by 16.6 °C (Fig. 1E). The L130R/G133D double mutantshowed an effect that was slightly less than additive, with a Tm thatwas increased by 20.9 °C compared with the WT protein (Fig. 1F).Whereas the ΔH values for the C-terminal domain were similar,the ΔH values for the N-terminal domain of the G133D single andL130R/G133D double mutant were considerably higher (97 and56 kcal/mol) than for the WT (12 kcal/mol; Table S1).

Structural Basis for Increased Stability of the FHbp Antigen. Toelucidate the structural basis for the increased stability of theFHbp ID 22 L130R/G133D double mutant, we crystallized theprotein and determined its X-ray structure to 1.6 Å resolution[Protein Data Bank (PDB) ID 4Z3T]. We solved the structureusing molecular replacement with a homology model of FHbpID 22 (21). The crystals were in space group P21 with two FHbpmolecules in the asymmetric unit. We refined the model to anRwork of 0.194 and an Rfree of 0.231 with good geometry; the dataand refinement statistics are shown in Table S2. The structure ofone of the two molecules (chain A) and the positions of the twosubstituted residues are shown schematically (Fig. 2A). Theelectron density map was interpretable for residues 7–255 inchain A and 12–23 and 31–254 in chain B. Based on a super-position of the two chains, the rms deviation was 0.20 Å, and thelargest backbone differences were 1–2 Å in the regions of resi-dues 140 and 200, which were sites of crystal contacts.There was clear electron density in the region of the two

substituted residues (Fig. 2B). The amino acid substitutions wereconfirmed by calculating a simulated annealing omit map from amodel excluding the side chains of residues 130 and 133 (Fig.2C). Residue R130 mediated ionic and H-bonding interactionswith R80 and E92 (Fig. 2D). Residue D133 formed hydrogenbonds with three backbone nitrogen atoms (residues 37–39) anda fourth H bond with the side-chain hydroxyl group of S39 (Fig.2E). Both sets of interactions anchored regions that were distantin the primary sequence, which collectively served to stabilize theN-terminal domain.

Stabilization of Epitopes Recognized by Protective Antibodies. Toinvestigate whether the increased thermal stability of the N-terminaldomain of the double-mutant FHbp translated into epitopes rec-ognized by anti-FHbp antibodies being stabilized in the doublemutant protein, we measured the binding of three murine anti-FHbp MAbs to FHbp by ELISA. MAb JAR 4, which recognizes anepitope in the N-terminal domain of FHbp (22), bound ∼2.6-foldbetter to the double mutant than to the WT protein (i.e., theconcentration of MAb needed to obtain an optical density of 2.0was 2.6-fold lower for the mutant than for the WT; Fig. 3A).Similarly, MAb JAR 41, which also recognizes an epitope inthe N-terminal domain of FHbp (23), bound approximately

Fig. 1. Thermal stability of FHbp mutants. (A) FHbp ID 1 WT (solid blackline) and R130L mutant (dashed gray line). (B) D133G mutant. (C) R130L/D133G double mutant. The data for ID 1 WT are the same as in Fig. 1A. TheR130L, D133G, and R130L/D133G mutants have transition midpoint (Tm)values for the N-terminal domain that are 4, 10, and 16 °C lower than that ofthe WT protein (Table S1). (D) FHbp ID 22 WT (solid black line) and L130Rsingle mutant (dashed gray line). (E) G133D mutant. (F) L130R/G133D doublemutant. The data for ID 22 WT are the same as in Fig. 1B. The L130R, G133D,and L130R/G133D mutants have transition midpoint (Tm) values for theN-terminal domain that are 8, 17, and 21 °C higher than that of the WT protein(Table S1). Data are shown for one of two to three experiments.

14824 | www.pnas.org/cgi/doi/10.1073/pnas.1507829112 Konar et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 7

, 202

0

threefold better to the double mutant (Fig. 3B). Binding of athird MAb, JAR 31, which recognizes an epitope that is at leastpartially in the C-terminal domain of FHbp, was similar forthe WT and double mutant (Fig. 3C). Thus, the stabilizationof epitopes appeared to be specific for those residing in theN-terminal domain.

To assess whether the stabilization of epitopes was due toincreased affinity of the MAbs for FHbp, we also measuredbinding of the MAbs to FHbp by surface plasmon resonance(SPR). MAbs JAR 4 and JAR 41 had affinities for the doublemutant that were 2.4- and 1.8-fold higher than for the WTFHbp. The dissociation constants (KD) were significantly dif-ferent (P ≤ 0.004; Table 1). In contrast, MAb JAR 31 had similaraffinity for the WT and double mutant (P = 0.26). The higheraffinities for the double-mutant protein partially resulted fromdifferences in the off-rates for antibody binding (kd; Table 1).Binding of human FH decreases FHbp immunogenicity (24–26),

and FHbp mutants with decreased binding of FH elicit higherprotective antibody responses in human FH transgenic mice (16,24, 27). Because R130 is an FH binding residue in variant group1 proteins (28), we tested binding of human FH to the double-mutant FHbp. By ELISA, FH bound ∼17-fold less to the doublemutant than to the WT protein (Fig. S3A). A control Penta-HisMAb, which was specific for the C-terminal His6 tag, boundsimilarly to the two recombinant proteins (Fig. S3B). We furthercharacterized the affinity of FH for FHbp by SPR by capturingFH and injecting FHbp as the analyte. The ID 22 WT had adissociation constant (KD) of 17.7 ± 1.2 nM, whereas the doublemutant had a KD of 118.0 ± 10.5 nM (P = 0.0007). Represen-tative sensorgrams are shown in Fig. S3 C and D.To determine whether the FHbp epitopes were stabilized

in vivo, we constructed isogenic mutant meningococcal strainsexpressing an FHbp ID 22 WT or double-mutant protein on theirsurface. By flow cytometry, MAb JAR 4 showed significantlyhigher binding to the strain with the double mutant compared withthe strain with the WT protein. The double mutant had a meanmedian fluorescence intensity (MFI) of 926 ± 59, compared withthe WT with an MFI of 398 ± 13 (Fig. 4A). Binding of MAbs JAR41 and JAR 31 showed more subtle but reproducibly higherbinding to the double mutant (Fig. 4 B and C). As a control, wetested a MAb that recognizes an unrelated surface protein, PorA,which bound similarly to the two strains (Fig. 4D). Calculation ofthe means and SE of triplicate measurements indicated that thedifferences in binding of the anti-FHbp MAbs to the WT anddouble mutant were significant (P ≤ 0.0015; Fig. 4 E–G) andparalleled those measured to the recombinant protein by SPR.

DiscussionMeningococcal FHbp vaccine antigens in variant groups 1 and 3were reported to be stable (15, 17, 18), whereas FHbp in variantgroup 2 was unstable (15, 16). In the present study, we usedmultiple sequence alignments and a known crystal structure ofFHbp in variant group 1 (19) to identify L130 and G133 as residuesthat might explain the lower thermal stability of FHbp variant group2. The crystal structure of FHbp (ID 1 in variant group 1) contains asalt bridge network that includes residues E44, R80, E92, and R130,and replacement of any of these residues with alanine decreases the

Fig. 2. Crystal structure of FHbp double mutant determined to 1.6 Å res-olution. (A) Position of substituted residues R130 and D133 in a view of theentire structure. The N-terminal domain is on the left, and the C-terminaldomain is on the right (the termini are on the back surface of the protein).(B) Electron density map in the region of substituted residues R130 andD133. The map (2mFo-DFc) is contoured at 1.2 σ. (C) Simulated annealingomit map of substituted residues R130 and D133. The side chain atoms wereomitted from the map calculation, and the difference map (mFo-DFc) con-toured at 3 σ is shown in purple. (D) Charged interaction network mediatedby substituted residue R130. One salt bridge and one H bond are formedbetween R130 and E92; another H bond is formed between R130 and R80.The distances between nonhydrogen atoms are shown in Å. (E) Charged Hbonds formed by substituted residue D133. Three H bonds are formed withbackbone N atoms (residues 37–39) and an additional H bond with the sidechain hydroxyl of S39 (shown in two alternative conformations). The dis-tances between nonhydrogen atoms are shown in Ångstroms. The imagewas generated using PyMOL (Version 1.3 Schrödinger, LLC).

Fig. 3. Binding of anti-FHbp MAbs to FHbp ID 22 by ELISA. (A) Binding of MAb JAR 4, which recognizes an epitope in the N-terminal domain of FHbp (22).A shift to the left (i.e., lower concentration) for the L130R/G133D mutant indicates higher binding compared with the WT. (B) Binding of MAb JAR 41, whichrecognizes an epitope in the N-terminal domain of FHbp (23). (C) Binding of control MAb JAR 31, which recognizes an epitope in the C-terminal domain ofFHbp. The data shown are the means and 2 SE of six replicates derived from two experiments.

Konar et al. PNAS | December 1, 2015 | vol. 112 | no. 48 | 14825

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Mar

ch 7

, 202

0

stability of the N-terminal domain. In contrast, the variant group 3proteins contain polar and hydrophobic residues at the corre-sponding positions (G44, Q80, N92, and L130). The variant group 2proteins are natural chimeras, with residues 1–93 derived from avariant group 1 lineage and residues 94–255 derived from avariant group 3 lineage (29). The chimeric nature of the N-ter-minal domain (residues 1–136) in variant group 2 proteins ap-pears to compromise its stability. Residues 130 and 133 reside atthe interface between the two portions of this hybrid domain(Fig. S4A). The low thermal stability for the N-terminal domainof FHbp in variant group 2 was suggested to limit its usefulnessas a vaccine antigen (15).

To overcome this potential limitation, we engineered an FHbpantigen in variant group 2 with substantially increased thermalstability and determined its crystal structure at high resolution.To date, crystal structures have been reported for FHbp fromvariant group 1 alone (19) and in a complex with a fragment ofhuman FH (28) and FHbp from variant group 3 in a complexwith the FH fragment (15). Previous attempts to crystallize a WTFHbp variant group 2 protein resulted in the structure of a pro-teolytic fragment containing only the C-terminal domain (15).Thus, the structure of the double mutant FHbp reported here isthe first structure of an intact FHbp antigen in variant group 2.Overall, the structure of the double mutant showed similarity to

Table 1. Kinetic parameters for binding of MAbs to FHbp WT or double mutant (DM) by SPR*

FHbp ID 22 MAb KD × 109† ka × 10−5‡ kd × 104§ Rmax{ χ2# P value for KD

jj

WT JAR 4 278 ± 34 1.06 ± 0.13 289 ± 19 106 ± 9 0.6 ± 0.2 0.001DM JAR 4 115 ± 9 1.56 ± 0.19 185 ± 37 138 ± 1 7.4 ± 3.3WT JAR 41 24.9 ± 1.6 1.23 ± 0.07 3.3 ± 0.2 77 ± 1 0.4 ± 0.1 0.004DM JAR 41 13.8 ± 0.9 2.01 ± 0.15 2.74 ± 0.06 119 ± 28 5.5 ± 2.6WT JAR 31 6.3 ± 0.4 1.43 ± 0.09 6.4 ± 0.3 201 ± 41 0.9 ± 0.3 0.26DM JAR 31 4.84 ± 0.51 1.44 ± 0.09 6.82 ± 0.32 219 ± 43 0.7 ± 0.0

*Mean ± SE from four to five replicate experiments.†KD, dissociation constant calculated as kd/ka (M); relative affinities cited in text from KD ratio (L130R/G133Dmutant/WT).‡ka, association rate constant (M–1·s–1).§kd, dissociation rate constant (s−1).{Rmax, maximal response units (RU).#χ2, goodness of fit to 1:1 binding model (RU2).jjP value from unpaired, parametric t test (two-tailed).

Fig. 4. Binding of anti-FHbp MAbs to the surface of meningococci. Isogenic mutant strains expressed either FHbp ID 22 WT or L130R/G133D double mutant.(A) Binding of JAR 4, which binds to an epitope in the N-terminal domain of FHbp (22). WT, solid line; double mutant, dashed line; negative control (noprimary antibody), gray shaded histogram. The means and the 2 SE of the MFI values are shown next to each histogram (MFI for the negative control was12.5). (B) Binding of JAR 41, which binds to an epitope in the N-terminal domain (23). (C) Binding of JAR 31. (D) Binding of a control MAb to a control protein,PorA. (E) MFI of individual measurements for MAb JAR 4 performed in triplicate. The horizontal bar represents the mean, and the error bars represent the2 SE values. (F) MAb JAR 41. (G) MAb JAR 31. (H) Control anti-PorA P1.7 MAb.

14826 | www.pnas.org/cgi/doi/10.1073/pnas.1507829112 Konar et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 7

, 202

0

that of FHbp ID 1 (19), with a rms deviation for backbone atomsof 0.46 Å (Fig. S4B). The main differences in backbone confor-mations were in the 80s loop (∼7 Å shift), the 210s loop (∼3 Å),and three other peripheral loops (∼2 Å), all of which were regionsinvolved in contacts between adjacent molecules in the crystal lattice.In the crystal structure of the FHbp double mutant, one of the

two substituted residues, R130, was involved in a salt bridgenetwork with E44, R80, and E92 that was similar to the networkin FHbp ID 1 (Fig. S4C); however, there were slight differencesin the side chain rotameric conformations of E92 and R130. Thesecond substituted residue, D133, exhibited similar conforma-tions in the two structures. Thus, by substitution of two aminoacid residues in ID 22 with those present in ID 1, we recreatednearly the same atomic interactions and increased the thermalstability by 21 °C. This was possible despite considerable aminoacid sequence divergence between the two proteins (69% over-all) and in particular in the region flanking 130–133 (Fig. S2).FHbp mutants with decreased binding of human FH were shown

to have enhanced immunogenicity in human FH transgenic mice(15, 16, 24, 27). The stabilized double-mutant FHbp had approxi-mately sevenfold lower affinity for human FH than the WT FHbp(Fig. S3). Residue R130 is known to stabilize the interaction ofFHbp ID 1 with human FH (15, 28); therefore, we did not expectthat the introduction of R130 in ID 22 would destabilize the in-teraction with FH. In the structure of FHbp ID 1 in a complex witha fragment of human FH, FHbp residue R130 mediates H bondswith two FH tyrosine hydroxyl groups, whereas D133 is ∼7 Å fromthe closest atoms in FH (Fig. S4D). Because R130 and D133 adoptsimilar side chain rotameric conformations in the structures of thedouble-mutant FHbp and the FHbp–FH fragment complex, thedecrease in affinity for FH in the double mutant may be due tosubtle structural changes near the mutation sites. The observa-tion of decreased affinity for the mutant despite having residuesR130 and D133 (as in FHbp ID 1 in variant group 1) supportsprevious data indicating that FHbps from different variant groupshave different binding modes for FH (15, 21).A comparison of the structures of a fragment of human FH in

complexes with FHbp ID 1 in variant group 1 (28) and ID 28 invariant group 3 (15) indicates that the FH interaction surfaces arein similar locations but that the residues involved differ. Further,FH has different affinities for FHbp from each of the three variantgroups as measured by SPR, and through analysis of site-specificmutants, the binding hotspots differ (15). Although a sevenfolddecrease in FH binding for the stabilized mutant might not besufficient for an optimal increase in FHbp immunogenicity in thepresence of human FH, in future studies additional amino acidsubstitutions (e.g., D211A, T221A, or others) (15, 16, 21) couldbe incorporated to obtain stable mutants with very low FH bindingand optimal immunogenicity. This approach has the furtheradvantage that the stabilizing substitutions might compensate forany possible effects of the substitutions that decrease FH bindingon destabilizing protective epitopes.As described in the opening paragraphs of this article, an in-

vestigational meningococcal vaccine contains a fusion of FHbpmolecules in variant groups 1–3 (11). A second experimental vac-cine employs native outer membrane vesicles that contain twooverexpressed FHbp variants, one each from variant groups 1and 2 (30). All three variant group 2 proteins tested to date havelow thermal stability compared with proteins in variant groups 1and 3 (Fig. S1) (15). The same amino acid substitutions that wereintroduced to stabilize FHbp ID 22 might also increase the stabilityof other FHbp variant group 2 proteins, which exist in two differentmodular groups based on the chimeric structure noted above (29).Using three complementary approaches, we demonstrated that

the stabilized double-mutant FHbp had higher binding of twocross-reactive MAbs. In ELISA, SPR, and flow cytometry experi-ments, MAbs JAR 4 and JAR 41 exhibited significantly higherbinding to the double mutant than to the WT FHbp (Figs. 3 and 4

and Table 1). These experiments highlight the conformationalintegrity of the stabilized double mutant, both in the form of therecombinant protein purified from Escherichia coli as well as thenative, lipidated protein expressed on the surface of meningoc-cocci. Because residues involved in the epitopes recognized byMAbs JAR 4 and JAR 41 are located outside the FH binding siteand neither of these MAbs inhibits binding of FH to FHbp, theeffect of the amino acid substitutions on MAb epitopes appears tobe indirect via stabilization of the N-terminal structural domain.The unfolding of the N-terminal domain of FHbp and several

of the variant group 2 proteins tested occurs at temperatures(38.5–40.6 °C) that are only slightly higher than the physiologicaltemperature, and therefore, some FHbp epitopes may be labileat this temperature. A stabilized variant group 2 protein is predictedto present epitopes and elicit protective antibody responses that areas high or higher than the WT protein. The stabilized mutant alsomight facilitate the production and formulation of vaccines con-taining certain recombinant FHbp variants. Finally, a stabilizedFHbp mutant might have increased surface expression and/or betterpresentation of epitopes in meningococci, which could enhance theimmunogenicity of native outer membrane vesicle vaccines.

Materials and MethodsSite-Specific Mutagenesis. Directed mutagenesis was carried out using thepET21b-FHbp ID 1 (4) or ID 22 (6) plasmids as templates. PCR amplificationand ligation was carried out using the procedures described in the PhusionSite-Directed Mutagenesis Kit (Thermo Scientific). The sequences of the oli-gonucleotide primers are listed in Table S3.

Protein Expression and Purification. FHbp was expressed from pET21-basedplasmids in E. coli strain BL21(DE3). Lysis was performed with three freeze/thawcycles in the presence of 0.1 mg/mL each of lysozyme, DNase (Sigma), andRNase A (Qiagen) in 50 mM NaPO4, 500 mM NaCl, 20 mM imidazole pH 7.4.The clarified lysate was loaded on a HiTrap Chelating HP column (5 mL, GELife Sciences), and FHbp was eluted with a linear gradient of imidazole.Fractions containing FHbp were dialyzed against 25 mM 2(N-morpholino)ethanesulfonic acid, 150 mM NaCl, pH 5.5, and loaded on a HiTrap SP HP column(5 mL, GE Life Sciences). FHbp was eluted with a linear gradient to 750 mMNaCl. The protein concentration was measured by absorbance at 280 nm(Nanodrop 1000, Thermo Scientific), using the molar extinction coefficientcalculated from the amino acid sequence using ProtParam (31).

ELISA. Binding of human FH and anti-FHbp MAbs to purified FHbp wasperformed as described (9). Briefly the wells of a microtiter plate werecoated with 2 μg/mL of purified recombinant FHbp in PBS. After blockingwith PBS containing 0.01% (wt/vol) sodium azide and 1% (wt/vol) BSA, serialdilutions of purified human FH, anti-FHbp MAbs, or a control Penta-His MAb(Qiagen) were added and incubated at room temperature for 1 h. Bound FHwas detected with sheep anti-FH (1:7,000; Abcam) and donkey anti-sheepIgG conjugated to alkaline phosphatase (1:5,000; Sigma). Bound MAbs weredetected with goat anti-mouse IgG conjugated to alkaline phosphatase(1:5,000; Sigma). The ELISAs were developed using paranitrophenyl phos-phate substrate (1 mg/mL; Sigma), and the optical density at 405 nm (OD405 nm)was measured after 30 min at room temperature. Relative binding wascalculated as the ratio in the concentration of FH or MAb needed to obtainan OD405 nm of 2.0 for WT versus mutant FHbp. FH and mAb binding wastested in at least three experiments, and the means and 2 SE are shown fromtwo experiments, each performed in triplicate.

Differential Scanning Calorimetry. Thermal stability of FHbp was measuredusing a VP-DSC instrument (Microcal). The protein was dialyzed against PBS,and dialysis buffer was used as the reference solution. The protein concen-tration was 0.5 mg/mL, the scan rate was 60 °C/h, and the passive feedbackmode was used. The data were analyzed using a non-two state model usingOrigin 5.0 software (Microcal). Representative data are shown from two tothree experiments. Transition midpoint (Tm) and enthalpy change (ΔH) val-ues and associated errors reported are derived from curve fitting to the datafrom one experiment.

Crystallization and Structure Determination. FHbp ID 22 double-mutant pro-tein was dialyzed against 10 mM Tris·Cl, 25 mM NaCl, pH 7.0, and concen-trated to 9.7 mg/mL. Crystals were grown by vapor diffusion in hanging

Konar et al. PNAS | December 1, 2015 | vol. 112 | no. 48 | 14827

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Mar

ch 7

, 202

0

drops in 27% (wt/vol) PEG 3,350, 0.1 M Tris·Cl, pH 9.5, and 0.25 M MgCl2.X-ray diffraction data were collected on beamline 5.0.1 at the Advanced LightSource, Lawrence Berkeley National Laboratory (λ = 0.977 Å, 100 K). Datawere reduced with Mosflm/Scala, and the structure was solved by molecularreplacement using MolRep (32) and a homology model of FHbp ID 22 (21).The structure was refined using Phenix (33); noncrystallographic symmetryrestraints were used in the earlier rounds of refinement. Model building wasperformed with Coot (34). The refined structure contained backbone dihedralangles that were 97% in favored, 3% in allowed, and 0% in outlier regions. Theatomic coordinates and reflection data are deposited in PDB (ID code 4Z3T).

SPR. Kinetic parameters for binding of anti-FHbp MAbs to FHbp were de-termined using a Biacore X-100 Plus instrument (GE Life Sciences). We immobi-lized 14,000 response units (RUs) of goat anti-mouse IgG Fc-specific antibody(Jackson ImmunoResearch) on a CM5 biosensor chip with an Amine Coupling kit(GE Life Sciences). For single-cycle kinetics, ∼500 RUs of murine anti-FHbp MAbswere captured, and five concentrations (1–100 nM or 3.16–316 nM) of FHbpanalyte were injected. The MAbs used were JAR 4 (22, 35), JAR 41 (23), and JAR31 (6). For measuring affinity of FH for FHbp, a goat anti-human FH antibody(Complement Technologies) that had been affinity purified was amine coupledto a CM5 chip. This antibody was used to capture purified human FH, and FHbpanalyte was injected at concentrations from 1 to 100 nM. Data were analyzedwith a 1:1 binding model using Biacore X100 Evaluation software. Each bindingassay was performed in at least two duplicate or triplicate experiments, and themeans and SE of three to four measurements are given.

Detection of Mutant FHbp on the Surface of Meningococci. Two isogenicmutant meningococcal strains were constructed from strain H44/76. Thenative FHbp ID 1 genewas replaced by homologous recombinationwith FHbpID 22 (WT or double mutant) adjacent to an ermC cassette (25, 36). Eryth-romycin-resistant mutants were selected and verified by DNA sequencing ofthe fhbp gene. Bacteria were grown in Frantz medium (37) to an OD620 nm

of 0.6 and washed in Dulbecco’s PBS (Corning CellGro) containing 1%(wt/vol) BSA (Equitech Bio). FHbp was detected with murine MAbs to FHbp(see SPR) or PorA P1.7 (MN14C11.6) (38), followed by goat anti-mouse IgGconjugated to AlexaFluor 488 (Life Technologies). The fluorescence was readon a Fortessa flow cytometer (BD Biosciences), and the data were analyzedwith FlowJo X software (TreeStar). Binding was tested in three experiments,and data are shown from an experiment performed in triplicate.

Statistical Analyses. The means and SEs of ELISA and SPR and flow cytometrydata were calculated with Prism 6.0f (GraphPad) from three to six replicates.Statistical tests were performed using parametric t tests using Prism. Two-tailed P values ≤ 0.05 were considered to be statistically significant.

ACKNOWLEDGMENTS. We thank Charlotte Rosenfield and Andrew Fergusfor technical assistance. Research reported in this publication was supportedby National Institute of Allergy and Infectious Diseases of the NIH Grant R01AI 099125 (to P.T.B.). This work was also supported by the National Heart,Lung and Blood Institute of the NIH Award R25 HL 096365. The AdvancedLight Source is supported by the Director, Office of Science, Office of BasicEnergy Sciences, US Department of Energy under Contract DE-AC02-05CH11231.

1. Madico G, et al. (2006) The meningococcal vaccine candidate GNA1870 binds thecomplement regulatory protein factor H and enhances serum resistance. J Immunol177(1):501–510.

2. Brehony C, Wilson DJ, Maiden MC (2009) Variation of the factor H-binding protein ofNeisseria meningitidis. Microbiology 155(Pt 12):4155–4169.

3. Fletcher LD, et al. (2004) Vaccine potential of the Neisseria meningitidis 2086 lipo-protein. Infect Immun 72(4):2088–2100.

4. Masignani V, et al. (2003) Vaccination against Neisseria meningitidis using threevariants of the lipoprotein GNA1870. J Exp Med 197(6):789–799.

5. Jiang HQ, et al. (2010) Broad vaccine coverage predicted for a bivalent recombinantfactor H binding protein based vaccine to prevent serogroup B meningococcal dis-ease. Vaccine 28(37):6086–6093.

6. Pajon R, Beernink PT, Harrison LH, Granoff DM (2010) Frequency of factor H-bindingprotein modular groups and susceptibility to cross-reactive bactericidal activity ininvasive meningococcal isolates. Vaccine 28(9):2122–2129.

7. Findlow J, et al. (2010) Multicenter, open-label, randomized phase II controlled trial ofan investigational recombinant Meningococcal serogroup B vaccine with and withoutouter membrane vesicles, administered in infancy. Clin Infect Dis 51(10):1127–1137.

8. Snape MD, et al. (2010) Immunogenicity of two investigational serogroup B menin-gococcal vaccines in the first year of life: A randomized comparative trial. PediatrInfect Dis J 29(11):e71–e79.

9. Beernink PT, Granoff DM (2008) Bactericidal antibody responses induced by menin-gococcal recombinant chimeric factor H-binding protein vaccines. Infect Immun 76(6):2568–2575.

10. Scarselli M, et al. (2011) Rational design of a meningococcal antigen inducing broadprotective immunity. Sci Transl Med 3(91):91ra62.

11. Malito E, et al. (2013) Defining a protective epitope on factor H binding protein, a keymeningococcal virulence factor and vaccine antigen. Proc Natl Acad Sci USA 110(9):3304–3309.

12. Murphy E, et al. (2009) Sequence diversity of the factor H binding protein vaccinecandidate in epidemiologically relevant strains of serogroup B Neisseria meningitidis.J Infect Dis 200(3):379–389.

13. Wang X, et al. (2011) Prevalence and genetic diversity of candidate vaccine antigensamong invasive Neisseria meningitidis isolates in the United States. Vaccine 29(29-30):4739–4744.

14. Pajon R, Fergus AM, Koeberling O, Caugant DA, Granoff DM (2011) Meningococcalfactor H binding proteins in epidemic strains from Africa: Implications for vaccinedevelopment. PLoS Negl Trop Dis 5(9):e1302.

15. Johnson S, et al. (2012) Design and evaluation of meningococcal vaccines throughstructure-based modification of host and pathogen molecules. PLoS Pathog 8(10):e1002981.

16. Rossi R, Granoff DM, Beernink PT (2013) Meningococcal factor H-binding proteinvaccines with decreased binding to human complement factor H have enhancedimmunogenicity in human factor H transgenic mice. Vaccine 31(46):5451–5457.

17. Mascioni A, et al. (2009) Structural basis for the immunogenic properties of themeningococcal vaccine candidate LP2086. J Biol Chem 284(13):8738–8746.

18. Beernink PT, Shaughnessy J, Ram S, Granoff DM (2010) Impaired immunogenicity of ameningococcal factor H-binding protein vaccine engineered to eliminate factorH binding. Clin Vaccine Immunol 17(7):1074–1078.

19. Cendron L, Veggi D, Girardi E, Zanotti G (2011) Structure of the uncomplexed Neis-seria meningitidis factor H-binding protein fHbp (rLP2086). Acta Crystallogr SectF Struct Biol Cryst Commun 67(Pt 5):531–535.

20. Beernink PT, Caugant DA, Welsch JA, Koeberling O, Granoff DM (2009) Meningo-coccal factor H-binding protein variants expressed by epidemic capsular group A,W-135, and X strains from Africa. J Infect Dis 199(9):1360–1368.

21. Pajon R, Beernink PT, Granoff DM (2012) Design of meningococcal factor H bindingprotein mutant vaccines that do not bind human complement factor H. Infect Immun80(8):2667–2677.

22. Beernink PT, LoPasso C, Angiolillo A, Felici F, Granoff D (2009) A region of theN-terminal domain of meningococcal factor H-binding protein that elicits bactericidalantibody across antigenic variant groups. Mol Immunol 46(8-9):1647–1653.

23. Vu DM, Pajon R, Reason DC, Granoff DM (2012) A broadly cross-reactive monoclonalantibody against an epitope on the N-terminus of meningococcal fHbp. Sci Rep 2:341.

24. Beernink PT, et al. (2011) A meningococcal factor H binding protein mutant thateliminates factor H binding enhances protective antibody responses to vaccination.J Immunol 186(6):3606–3614.

25. Costa I, Pajon R, Granoff DM (2014) Human factor H (FH) impairs protective menin-gococcal anti-FHbp antibody responses and the antibodies enhance FH binding.MBio5(5):e01625–e14.

26. Granoff DM, et al. (2015) Binding of complement factor H (FH) decreases protectiveanti-FH binding protein antibody responses of infant rhesus macaques immunizedwith a meningococcal serogroup B vaccine. J Infect Dis 212(5):784–792, 10.1093/infdis/jiv081.

27. Beernink PT, et al. (2012) The effect of human factor H on immunogenicity of me-ningococcal native outer membrane vesicle vaccines with over-expressed factorH binding protein. PLoS Pathog 8(5):e1002688.

28. Schneider MC, et al. (2009) Neisseria meningitidis recruits factor H using proteinmimicry of host carbohydrates. Nature 458(7240):890–893.

29. Beernink PT, Granoff DM (2009) The modular architecture of meningococcal factorH-binding protein. Microbiology 155(Pt 9):2873–2883.

30. Koeberling O, Giuntini S, Seubert A, Granoff DM (2009) Meningococcal outer mem-brane vesicle vaccines derived from mutant strains engineered to express factorH binding proteins from antigenic variant groups 1 and 2. Clin Vaccine Immunol 16(2):156–162.

31. Gasteiger E, et al. (2005) Protein Identification and Analysis Tools on the ExPASyServer, ed Walker JM (Humana Press, Clifton, NJ), pp 571–607.

32. Collaborative Computational Project, Number 4 (1994) The CCP4 suite: Programs forprotein crystallography. Acta Crystallogr D Biol Crystallogr 50(Pt 5):760–763.

33. Adams PD, et al. (2010) PHENIX: A comprehensive Python-based system for macro-molecular structure solution. Acta Crystallogr D Biol Crystallogr 66(Pt 2):213–221.

34. Emsley P, Lohkamp B, Scott WG, Cowtan K (2010) Features and development of Coot.Acta Crystallogr D Biol Crystallogr 66(Pt 4):486–501.

35. Welsch JA, Rossi R, Comanducci M, Granoff DM (2004) Protective activity of mono-clonal antibodies to genome-derived neisserial antigen 1870, a Neisseria meningitidiscandidate vaccine. J Immunol 172(9):5606–5615.

36. Konar M, Granoff DM, Beernink PT (2013) Importance of inhibition of binding ofcomplement factor H for serum bactericidal antibody responses to meningococcalfactor H-binding protein vaccines. J Infect Dis 208(4):627–636.

37. Frasch CE, van Alphen L, Holst J, Poolman JT, Rosenqvist E (2001) Outer membranevesicle vaccines for meningococcal disease. Meningococcal Vaccines: Methods andProtocols, eds Pollard AJ, Maiden MC (Humana Press, Totawa, NJ), pp 81–87.

38. Abdillahi H, Poolman JT (1988) Neisseria meningitidis group B serosubtyping usingmonoclonal antibodies in whole-cell ELISA. Microb Pathog 4(1):27–32.

39. Larkin MA, et al. (2007) Clustal W and Clustal X version 2.0. Bioinformatics 23(21):2947–2948.

14828 | www.pnas.org/cgi/doi/10.1073/pnas.1507829112 Konar et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 7

, 202

0